We propose a carefully integrated and theoretically-motivated set of behavioral and neuroimaging studies designed both to broaden and deepen current understanding of age-related changes in implicit memory. Most studies report that implicit memory is preserved over chronological age as well as across time within an age group, whereas explicit memory declines in elderly subjects. However, some provocative findings have been reported, showing implicit memory loss with certain types of stimuli, and over temporal delays. Using a possible/impossible object decision priming task, we found that decision accuracy was significantly facilitated or primed for studied, but not for non-studied possible objects in both young and elderly subjects. However, young subjects continued to show priming up to 1 week after study, whereas priming virtually disappeared for the elders after 20 minutes. This suggestion of a deteriorating perceptual memory system in the elderly constitutes a central theme in our proposed work. In particular, the hypothesis to be addressed is that familiar or pre-existing representations can be remembered implicitly by older individuals; whereas implicit memory for novel representations may deteriorate over time. Our strategy will be to establish behavioral methodology and findings with behavioral studies, and then to attempt to explore the neural substrates that support these tasks with followup H215O PET studies. Thus, beginning in year 1, we will conduct a series of 6 behavioral experiments designed to test our hypotheses, culminating in a study where we will use a training procedure to try to induce familiarity in our novel stimuli in order to directly improve delayed implicit memory performance in elders. Beginning in year 2, we will conduct a series of PET studies to follow up on the behavioral studies and assess the neural substrates supporting encoding and retrieval of implicit memory both immediately and at delays in young and older individuals. This proposal represents a unique combination of expertise across collaborators and will make a major contribution to our understanding of age-related changes in memory.